Material property estimation using nonlinear inversion
Abstract
A particularly stable form of nonlinear inversion known as Ridge Regression or Marquardt's method is implemented for material property estimation from a set of measured data. The technique permits inverse solutions to be obtained which are more stable than standard steepest descent or Newton-Gauss type algorithms. The implementation guarantees stability but not uniqueness and the solution is, among other things, a function of the initial guess model parameters. The inversion of a pulse echo time sequence is used as an illustrative example, illustrating the importance of parameter confidence intervals.
- Publication:
-
Review of Progress in Quantitative Nondestructive Evaluation. Vol. 9B
- Pub Date:
- 1990
- Bibcode:
- 1990qnde....9..101P
- Keywords:
-
- Materials Science;
- Mathematical Models;
- Mechanical Properties;
- Nonlinear Equations;
- Parameter Identification;
- Wave Equations;
- Correlation Coefficients;
- Data Acquisition;
- Error Analysis;
- Optimization;
- Sound Waves;
- Acoustics